Diagnostic Review revisions contained larger shares of meaning-change, organization, and content/argument work, while Control revisions remained more dependent on surface-level editing¶

Claim · All claims
Evidence · 1 study · causal r2 · q2
Currency · newest study 2026 · checked against its sources 2026-10-16 (automated, 1 of 1 entry) · not reviewed by a person

Subclaims¶

q2 i? The Control group's surface-level revisions accounted for 36.5%, 42.4%, and 40.6% of coded revisions across the three cycles, while Diagnostic Review surface revisions stayed near one-third with larger combined shares of meaning-change, organization, and content/argument revisions. → Yao X-C and Fan L 2026

Evidence¶

Yao X-C and Fan L 2026¶

Yao X-C and Fan L (2026) From automated scoring to learning diagnosis: a mechanism study of AI-supported formative assessment in English writing. Front. Psychol. 17:1905455. https://doi.org/10.3389/fpsyg.2026.1905455

q2 · i? · causal · r2

Revision coding across the three cycles in the two-class quasi-experiment, shown in stacked area plots with revision quality. "The Control group remained more dependent on surface-level revision", while Diagnostic Review revisions shifted toward meaning change, organization, and content/argument work.

"The Control group remained more dependent on surface-level revision, which accounted for 36.5%, 42.4%, and 40.6% of coded revisions across the three cycles. In the Diagnostic Review group, surface-level revisionsstayednearone-thirdofrevisions,whilemeaning-change, organization, and content/argument revisions formed a larger combinedshare."

Studies bearing on this claim¶

What leads learners to make deep, meaning-level revisions rather than surface edits? This table gathers the studies behind this claim and the 10 narrower claims that bear on it: 9 distinct studies (6 causal, 3 associational), q2–q3, 1 of 9 with an effect size coded. One row per study, codes as each entry records them; no effect is averaged and no verdict is drawn. Built by scripts/synthesize_claim.py from eval/synthesis/; a row that looks wrong is fixed on the claim it comes from.

Discussion¶

Learner Variables¶

• Writing Skills — outcome: instruction changes it

Related Claims¶

• AI-teacher diagnostic agreement was higher in the Diagnostic Review class and varied by issue type, with local language issues best diagnosed — related
• The Diagnostic Review class scored higher on diagnostic accuracy, actionability, comprehension, revision quality, and self-assessment accuracy across all three cycles — related
• The Diagnostic Review class gained more writing points across three cycles than the Control class (Hedges' g = 1.12) — related
• Diagnostic Review reflection records contained fewer misunderstanding codes and higher reflection quality than Control follow-up records — related
• Assessment for learning improves achievement — a broader claim this one bears on
• Peer Feedback Improves Writing — related
• High-acceptance groups made more effective use of AI feedback during revision, resulting in higher accuracy scores — related
• Teacher review workload was higher in the Diagnostic Review class in review minutes, added instruction, and review intensity — related
• FRAC and the combined intervention shifted revision toward deeper levels, with deep revision (L3+L4) rising from 25% in the regular AI group to 55% in the combined group — related